From bcb4f7e98c2f87b08866d940db5f0e6f71b2fac1 Mon Sep 17 00:00:00 2001
From: OscarXu <huaiguxu@amd.com>
Date: Thu, 15 May 2025 16:22:45 +0800
Subject: [PATCH] Add gemm1 v1 to selector

---
 ...ne_xdlops_b_preshuffle_mx_moe_selector.hpp | 24 ++++++++++++++++++-
 ...dlops_b_preshuflle_gufusion_v1_moe_mx.hpp} |  2 +-
 2 files changed, 24 insertions(+), 2 deletions(-)
 rename include/ck/tensor_operation/gpu/block/{blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_mx.hpp => blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_moe_mx.hpp} (99%)

diff --git a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_selector.hpp b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_selector.hpp
index 4eeb9ccc14..7fd8040543 100644
--- a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_selector.hpp
+++ b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_selector.hpp
@@ -4,6 +4,7 @@
 #pragma once
 
 #include "ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_v1_moe_mx.hpp"
+#include "ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_moe_mx.hpp"
 
 namespace ck {
 
@@ -71,7 +72,28 @@ constexpr auto BlockGemmMXBPreshufflePipeline_Selector()
     {
         if constexpr(GUFusion)
         {
-            return nullptr;
+            return return BlockwiseGemmXdlops_pipeline_bpreshuffle_gufusion_v1_moe_mx<
+                BlkGemmPipeSche,
+                ThreadBlockSize,
+                ScaleBlockSize,
+                ADataType,
+                AScaleDataType,
+                BDataType,
+                BScaleDataType,
+                ATileDesc,
+                BTileDesc,
+                AMmaTileDesc,
+                BMmaTileDesc,
+                ABlockTransferSrcScalarPerVector,
+                BBlockTransferSrcScalarPerVector,
+                MPerBlock,
+                NPerBlock,
+                KPerBlock,
+                MPerXDL,
+                NPerXDL,
+                MRepeat,
+                NRepeat,
+                KPack>{};;
         }
         else
         {
diff --git a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_mx.hpp b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_moe_mx.hpp
similarity index 99%
rename from include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_mx.hpp
rename to include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_moe_mx.hpp
index 0b4c6f10cb..641fafa3cb 100644
--- a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_mx.hpp
+++ b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_moe_mx.hpp
@@ -58,7 +58,7 @@ template <index_t ThreadBlockSize,
           index_t MRepeat, // MXdlPerWave
           index_t NRepeat, // NXdlPerWave
           index_t KPack>
-struct BlockwiseGemmXdlops_pipeline_bpreshuffle_gufusion_v1_mx_tmp<BlockGemmPipelineScheduler::Intrawave,
+struct BlockwiseGemmXdlops_pipeline_bpreshuffle_gufusion_v1_moe_mx<BlockGemmPipelineScheduler::Intrawave,
                                                           ThreadBlockSize,
                                                           ScaleBlockSize,
                                                           ADataType,